# Discovering Higgs Boson signal from background noise

import numpy as np

#loading training.csv to memory

#two unique values signifying signal and background noise

array(['s', 'b'], dtype=object)

#replacing string with numerical values to run numerical methods

df['Label'] = x

#converting label to int

df.drop('EventId', axis=1).describe()

import matplotlib.pyplot as plt

x

Text(0.5, 1, 'Correlation matrix')

#Robust scaler works best with outliers

scaler = RobustScaler()

#scaling all values except the index and target variables

RobustScaler(copy=True, quantile_range=(25.0, 75.0), with_centering=True,
             with_scaling=True)

scaled_data = scaler.transform(df.drop(['EventId', 'Weight', 'Label'], axis=1))

from sklearn.decomposition import PCA

#testing PCA with 2 components

pca.fit(scaled_data)

PCA(copy=True, iterated_power='auto', n_components=2, random_state=None,
    svd_solver='auto', tol=0.0, whiten=False)

pca_data = pca.transform(scaled_data)

#1st component of PCA covers 79.9% of the total variance

array([0.79941427, 0.07158938])

for i in range(0, 13):

pca.explained_variance_ratio_

array([0.79941427, 0.07158938, 0.02945643, 0.02295639, 0.01351438,
       0.00996207, 0.00880798, 0.0084082 , 0.00619115, 0.00534548,
       0.00445186, 0.00405008])

var_sum = []

[0.7994142682423108,
 0.8710036465167198,
 0.9004600789164949,
 0.9234164705934974,
 0.9369308505878781,
 0.946892924605498,
 0.9557009032955126,
 0.964109108282042,
 0.9703002544228476,
 0.9756457300864789,
 0.9800975889437665,
 0.9841476721518224]

plt.figure(figsize=(14,10))

[<matplotlib.lines.Line2D at 0x7f769315d630>]

pca.components_.shape

(12, 30)

#this dataframe is the correlation of each feature with each principal component

df_comp

plt.figure(figsize=(16,12))

<matplotlib.axes._subplots.AxesSubplot at 0x7f7693d795c0>

​

​

xxxxxxxxxx

​

​

from sklearn.manifold import TSNE

x

pca_data.shape

(250000, 12)

import time

#measuring elapsed time to transform tsne model

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 250000 samples in 0.285s...
[t-SNE] Computed neighbors for 250000 samples in 138.546s...
[t-SNE] Computed conditional probabilities for sample 1000 / 250000
[t-SNE] Computed conditional probabilities for sample 2000 / 250000
[t-SNE] Computed conditional probabilities for sample 3000 / 250000
[t-SNE] Computed conditional probabilities for sample 4000 / 250000
[t-SNE] Computed conditional probabilities for sample 5000 / 250000
[t-SNE] Computed conditional probabilities for sample 6000 / 250000
[t-SNE] Computed conditional probabilities for sample 7000 / 250000
[t-SNE] Computed conditional probabilities for sample 8000 / 250000
[t-SNE] Computed conditional probabilities for sample 9000 / 250000
[t-SNE] Computed conditional probabilities for sample 10000 / 250000
[t-SNE] Computed conditional probabilities for sample 11000 / 250000
[t-SNE] Computed conditional probabilities for sample 12000 / 250000
[t-SNE] Computed conditional probabilities for sample 13000 / 250000
[t-SNE] Computed conditional probabilities for sample 14000 / 250000
[t-SNE] Computed conditional probabilities for sample 15000 / 250000
[t-SNE] Computed conditional probabilities for sample 16000 / 250000
[t-SNE] Computed conditional probabilities for sample 17000 / 250000
[t-SNE] Computed conditional probabilities for sample 18000 / 250000
[t-SNE] Computed conditional probabilities for sample 19000 / 250000
[t-SNE] Computed conditional probabilities for sample 20000 / 250000
[t-SNE] Computed conditional probabilities for sample 21000 / 250000
[t-SNE] Computed conditional probabilities for sample 22000 / 250000
[t-SNE] Computed conditional probabilities for sample 23000 / 250000
[t-SNE] Computed conditional probabilities for sample 24000 / 250000
[t-SNE] Computed conditional probabilities for sample 25000 / 250000
[t-SNE] Computed conditional probabilities for sample 26000 / 250000
[t-SNE] Computed conditional probabilities for sample 27000 / 250000
[t-SNE] Computed conditional probabilities for sample 28000 / 250000
[t-SNE] Computed conditional probabilities for sample 29000 / 250000
[t-SNE] Computed conditional probabilities for sample 30000 / 250000
[t-SNE] Computed conditional probabilities for sample 31000 / 250000
[t-SNE] Computed conditional probabilities for sample 32000 / 250000
[t-SNE] Computed conditional probabilities for sample 33000 / 250000
[t-SNE] Computed conditional probabilities for sample 34000 / 250000
[t-SNE] Computed conditional probabilities for sample 35000 / 250000
[t-SNE] Computed conditional probabilities for sample 36000 / 250000
[t-SNE] Computed conditional probabilities for sample 37000 / 250000
[t-SNE] Computed conditional probabilities for sample 38000 / 250000
[t-SNE] Computed conditional probabilities for sample 39000 / 250000
[t-SNE] Computed conditional probabilities for sample 40000 / 250000
[t-SNE] Computed conditional probabilities for sample 41000 / 250000
[t-SNE] Computed conditional probabilities for sample 42000 / 250000
[t-SNE] Computed conditional probabilities for sample 43000 / 250000
[t-SNE] Computed conditional probabilities for sample 44000 / 250000
[t-SNE] Computed conditional probabilities for sample 45000 / 250000
[t-SNE] Computed conditional probabilities for sample 46000 / 250000
[t-SNE] Computed conditional probabilities for sample 47000 / 250000
[t-SNE] Computed conditional probabilities for sample 48000 / 250000
[t-SNE] Computed conditional probabilities for sample 49000 / 250000
[t-SNE] Computed conditional probabilities for sample 50000 / 250000
[t-SNE] Computed conditional probabilities for sample 51000 / 250000
[t-SNE] Computed conditional probabilities for sample 52000 / 250000
[t-SNE] Computed conditional probabilities for sample 53000 / 250000
[t-SNE] Computed conditional probabilities for sample 54000 / 250000
[t-SNE] Computed conditional probabilities for sample 55000 / 250000
[t-SNE] Computed conditional probabilities for sample 56000 / 250000
[t-SNE] Computed conditional probabilities for sample 57000 / 250000
[t-SNE] Computed conditional probabilities for sample 58000 / 250000
[t-SNE] Computed conditional probabilities for sample 59000 / 250000
[t-SNE] Computed conditional probabilities for sample 60000 / 250000
[t-SNE] Computed conditional probabilities for sample 61000 / 250000
[t-SNE] Computed conditional probabilities for sample 62000 / 250000
[t-SNE] Computed conditional probabilities for sample 63000 / 250000
[t-SNE] Computed conditional probabilities for sample 64000 / 250000
[t-SNE] Computed conditional probabilities for sample 65000 / 250000
[t-SNE] Computed conditional probabilities for sample 66000 / 250000
[t-SNE] Computed conditional probabilities for sample 67000 / 250000
[t-SNE] Computed conditional probabilities for sample 68000 / 250000
[t-SNE] Computed conditional probabilities for sample 69000 / 250000
[t-SNE] Computed conditional probabilities for sample 70000 / 250000
[t-SNE] Computed conditional probabilities for sample 71000 / 250000
[t-SNE] Computed conditional probabilities for sample 72000 / 250000
[t-SNE] Computed conditional probabilities for sample 73000 / 250000
[t-SNE] Computed conditional probabilities for sample 74000 / 250000
[t-SNE] Computed conditional probabilities for sample 75000 / 250000
[t-SNE] Computed conditional probabilities for sample 76000 / 250000
[t-SNE] Computed conditional probabilities for sample 77000 / 250000
[t-SNE] Computed conditional probabilities for sample 78000 / 250000
[t-SNE] Computed conditional probabilities for sample 79000 / 250000
[t-SNE] Computed conditional probabilities for sample 80000 / 250000
[t-SNE] Computed conditional probabilities for sample 81000 / 250000
[t-SNE] Computed conditional probabilities for sample 82000 / 250000
[t-SNE] Computed conditional probabilities for sample 83000 / 250000
[t-SNE] Computed conditional probabilities for sample 84000 / 250000
[t-SNE] Computed conditional probabilities for sample 85000 / 250000
[t-SNE] Computed conditional probabilities for sample 86000 / 250000
[t-SNE] Computed conditional probabilities for sample 87000 / 250000
[t-SNE] Computed conditional probabilities for sample 88000 / 250000
[t-SNE] Computed conditional probabilities for sample 89000 / 250000
[t-SNE] Computed conditional probabilities for sample 90000 / 250000
[t-SNE] Computed conditional probabilities for sample 91000 / 250000
[t-SNE] Computed conditional probabilities for sample 92000 / 250000
[t-SNE] Computed conditional probabilities for sample 93000 / 250000
[t-SNE] Computed conditional probabilities for sample 94000 / 250000
[t-SNE] Computed conditional probabilities for sample 95000 / 250000
[t-SNE] Computed conditional probabilities for sample 96000 / 250000
[t-SNE] Computed conditional probabilities for sample 97000 / 250000
[t-SNE] Computed conditional probabilities for sample 98000 / 250000
[t-SNE] Computed conditional probabilities for sample 99000 / 250000
[t-SNE] Computed conditional probabilities for sample 100000 / 250000
[t-SNE] Computed conditional probabilities for sample 101000 / 250000
[t-SNE] Computed conditional probabilities for sample 102000 / 250000
[t-SNE] Computed conditional probabilities for sample 103000 / 250000
[t-SNE] Computed conditional probabilities for sample 104000 / 250000
[t-SNE] Computed conditional probabilities for sample 105000 / 250000
[t-SNE] Computed conditional probabilities for sample 106000 / 250000
[t-SNE] Computed conditional probabilities for sample 107000 / 250000
[t-SNE] Computed conditional probabilities for sample 108000 / 250000
[t-SNE] Computed conditional probabilities for sample 109000 / 250000
[t-SNE] Computed conditional probabilities for sample 110000 / 250000
[t-SNE] Computed conditional probabilities for sample 111000 / 250000
[t-SNE] Computed conditional probabilities for sample 112000 / 250000
[t-SNE] Computed conditional probabilities for sample 113000 / 250000
[t-SNE] Computed conditional probabilities for sample 114000 / 250000
[t-SNE] Computed conditional probabilities for sample 115000 / 250000
[t-SNE] Computed conditional probabilities for sample 116000 / 250000
[t-SNE] Computed conditional probabilities for sample 117000 / 250000
[t-SNE] Computed conditional probabilities for sample 118000 / 250000
[t-SNE] Computed conditional probabilities for sample 119000 / 250000
[t-SNE] Computed conditional probabilities for sample 120000 / 250000
[t-SNE] Computed conditional probabilities for sample 121000 / 250000
[t-SNE] Computed conditional probabilities for sample 122000 / 250000
[t-SNE] Computed conditional probabilities for sample 123000 / 250000
[t-SNE] Computed conditional probabilities for sample 124000 / 250000
[t-SNE] Computed conditional probabilities for sample 125000 / 250000
[t-SNE] Computed conditional probabilities for sample 126000 / 250000

[t-SNE] Computed conditional probabilities for sample 127000 / 250000
[t-SNE] Computed conditional probabilities for sample 128000 / 250000
[t-SNE] Computed conditional probabilities for sample 129000 / 250000
[t-SNE] Computed conditional probabilities for sample 130000 / 250000
[t-SNE] Computed conditional probabilities for sample 131000 / 250000
[t-SNE] Computed conditional probabilities for sample 132000 / 250000
[t-SNE] Computed conditional probabilities for sample 133000 / 250000
[t-SNE] Computed conditional probabilities for sample 134000 / 250000
[t-SNE] Computed conditional probabilities for sample 135000 / 250000
[t-SNE] Computed conditional probabilities for sample 136000 / 250000
[t-SNE] Computed conditional probabilities for sample 137000 / 250000
[t-SNE] Computed conditional probabilities for sample 138000 / 250000
[t-SNE] Computed conditional probabilities for sample 139000 / 250000
[t-SNE] Computed conditional probabilities for sample 140000 / 250000
[t-SNE] Computed conditional probabilities for sample 141000 / 250000
[t-SNE] Computed conditional probabilities for sample 142000 / 250000
[t-SNE] Computed conditional probabilities for sample 143000 / 250000
[t-SNE] Computed conditional probabilities for sample 144000 / 250000
[t-SNE] Computed conditional probabilities for sample 145000 / 250000
[t-SNE] Computed conditional probabilities for sample 146000 / 250000
[t-SNE] Computed conditional probabilities for sample 147000 / 250000
[t-SNE] Computed conditional probabilities for sample 148000 / 250000
[t-SNE] Computed conditional probabilities for sample 149000 / 250000
[t-SNE] Computed conditional probabilities for sample 150000 / 250000
[t-SNE] Computed conditional probabilities for sample 151000 / 250000
[t-SNE] Computed conditional probabilities for sample 152000 / 250000
[t-SNE] Computed conditional probabilities for sample 153000 / 250000
[t-SNE] Computed conditional probabilities for sample 154000 / 250000
[t-SNE] Computed conditional probabilities for sample 155000 / 250000
[t-SNE] Computed conditional probabilities for sample 156000 / 250000
[t-SNE] Computed conditional probabilities for sample 157000 / 250000
[t-SNE] Computed conditional probabilities for sample 158000 / 250000
[t-SNE] Computed conditional probabilities for sample 159000 / 250000
[t-SNE] Computed conditional probabilities for sample 160000 / 250000
[t-SNE] Computed conditional probabilities for sample 161000 / 250000
[t-SNE] Computed conditional probabilities for sample 162000 / 250000
[t-SNE] Computed conditional probabilities for sample 163000 / 250000
[t-SNE] Computed conditional probabilities for sample 164000 / 250000
[t-SNE] Computed conditional probabilities for sample 165000 / 250000
[t-SNE] Computed conditional probabilities for sample 166000 / 250000
[t-SNE] Computed conditional probabilities for sample 167000 / 250000
[t-SNE] Computed conditional probabilities for sample 168000 / 250000
[t-SNE] Computed conditional probabilities for sample 169000 / 250000
[t-SNE] Computed conditional probabilities for sample 170000 / 250000
[t-SNE] Computed conditional probabilities for sample 171000 / 250000
[t-SNE] Computed conditional probabilities for sample 172000 / 250000
[t-SNE] Computed conditional probabilities for sample 173000 / 250000
[t-SNE] Computed conditional probabilities for sample 174000 / 250000
[t-SNE] Computed conditional probabilities for sample 175000 / 250000
[t-SNE] Computed conditional probabilities for sample 176000 / 250000
[t-SNE] Computed conditional probabilities for sample 177000 / 250000
[t-SNE] Computed conditional probabilities for sample 178000 / 250000
[t-SNE] Computed conditional probabilities for sample 179000 / 250000
[t-SNE] Computed conditional probabilities for sample 180000 / 250000
[t-SNE] Computed conditional probabilities for sample 181000 / 250000
[t-SNE] Computed conditional probabilities for sample 182000 / 250000
[t-SNE] Computed conditional probabilities for sample 183000 / 250000
[t-SNE] Computed conditional probabilities for sample 184000 / 250000
[t-SNE] Computed conditional probabilities for sample 185000 / 250000
[t-SNE] Computed conditional probabilities for sample 186000 / 250000
[t-SNE] Computed conditional probabilities for sample 187000 / 250000
[t-SNE] Computed conditional probabilities for sample 188000 / 250000
[t-SNE] Computed conditional probabilities for sample 189000 / 250000
[t-SNE] Computed conditional probabilities for sample 190000 / 250000
[t-SNE] Computed conditional probabilities for sample 191000 / 250000
[t-SNE] Computed conditional probabilities for sample 192000 / 250000
[t-SNE] Computed conditional probabilities for sample 193000 / 250000
[t-SNE] Computed conditional probabilities for sample 194000 / 250000
[t-SNE] Computed conditional probabilities for sample 195000 / 250000
[t-SNE] Computed conditional probabilities for sample 196000 / 250000
[t-SNE] Computed conditional probabilities for sample 197000 / 250000
[t-SNE] Computed conditional probabilities for sample 198000 / 250000
[t-SNE] Computed conditional probabilities for sample 199000 / 250000
[t-SNE] Computed conditional probabilities for sample 200000 / 250000
[t-SNE] Computed conditional probabilities for sample 201000 / 250000
[t-SNE] Computed conditional probabilities for sample 202000 / 250000
[t-SNE] Computed conditional probabilities for sample 203000 / 250000
[t-SNE] Computed conditional probabilities for sample 204000 / 250000
[t-SNE] Computed conditional probabilities for sample 205000 / 250000
[t-SNE] Computed conditional probabilities for sample 206000 / 250000
[t-SNE] Computed conditional probabilities for sample 207000 / 250000
[t-SNE] Computed conditional probabilities for sample 208000 / 250000
[t-SNE] Computed conditional probabilities for sample 209000 / 250000
[t-SNE] Computed conditional probabilities for sample 210000 / 250000
[t-SNE] Computed conditional probabilities for sample 211000 / 250000
[t-SNE] Computed conditional probabilities for sample 212000 / 250000
[t-SNE] Computed conditional probabilities for sample 213000 / 250000
[t-SNE] Computed conditional probabilities for sample 214000 / 250000
[t-SNE] Computed conditional probabilities for sample 215000 / 250000
[t-SNE] Computed conditional probabilities for sample 216000 / 250000
[t-SNE] Computed conditional probabilities for sample 217000 / 250000
[t-SNE] Computed conditional probabilities for sample 218000 / 250000
[t-SNE] Computed conditional probabilities for sample 219000 / 250000
[t-SNE] Computed conditional probabilities for sample 220000 / 250000
[t-SNE] Computed conditional probabilities for sample 221000 / 250000
[t-SNE] Computed conditional probabilities for sample 222000 / 250000
[t-SNE] Computed conditional probabilities for sample 223000 / 250000
[t-SNE] Computed conditional probabilities for sample 224000 / 250000
[t-SNE] Computed conditional probabilities for sample 225000 / 250000
[t-SNE] Computed conditional probabilities for sample 226000 / 250000
[t-SNE] Computed conditional probabilities for sample 227000 / 250000
[t-SNE] Computed conditional probabilities for sample 228000 / 250000
[t-SNE] Computed conditional probabilities for sample 229000 / 250000
[t-SNE] Computed conditional probabilities for sample 230000 / 250000
[t-SNE] Computed conditional probabilities for sample 231000 / 250000
[t-SNE] Computed conditional probabilities for sample 232000 / 250000
[t-SNE] Computed conditional probabilities for sample 233000 / 250000
[t-SNE] Computed conditional probabilities for sample 234000 / 250000
[t-SNE] Computed conditional probabilities for sample 235000 / 250000
[t-SNE] Computed conditional probabilities for sample 236000 / 250000
[t-SNE] Computed conditional probabilities for sample 237000 / 250000
[t-SNE] Computed conditional probabilities for sample 238000 / 250000
[t-SNE] Computed conditional probabilities for sample 239000 / 250000
[t-SNE] Computed conditional probabilities for sample 240000 / 250000
[t-SNE] Computed conditional probabilities for sample 241000 / 250000
[t-SNE] Computed conditional probabilities for sample 242000 / 250000
[t-SNE] Computed conditional probabilities for sample 243000 / 250000
[t-SNE] Computed conditional probabilities for sample 244000 / 250000
[t-SNE] Computed conditional probabilities for sample 245000 / 250000
[t-SNE] Computed conditional probabilities for sample 246000 / 250000
[t-SNE] Computed conditional probabilities for sample 247000 / 250000
[t-SNE] Computed conditional probabilities for sample 248000 / 250000
[t-SNE] Computed conditional probabilities for sample 249000 / 250000
[t-SNE] Computed conditional probabilities for sample 250000 / 250000
[t-SNE] Mean sigma: 0.387824

[t-SNE] KL divergence after 50 iterations with early exaggeration: 125.084518
[t-SNE] KL divergence after 250 iterations: 5.860478
t-SNE done! Time elapsed: 6838.500510692596 seconds

tsne_arr.shape

(250000, 3)

#creating a dataframe with 3 new components obtained by feature extraction using tsne

import plotly.express as px

tsne_df = pd.read_csv('tsne_data.csv')

fig = px.scatter_3d(tsne_df, x='tsne_1', y='tsne_2', z='tsne_3', color='label')

tsne_df.to_csv('tsne.csv', index=False)

#using saved results of t-SNE with 2 components

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f7603333320>

#applying t-SNE with 2 components on PCA-applied data

tsne1 = TSNE(n_components=2, n_iter=250, verbose=1, n_iter_without_progress=50)

time_start = time.time()

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 250000 samples in 0.282s...
[t-SNE] Computed neighbors for 250000 samples in 139.670s...
[t-SNE] Computed conditional probabilities for sample 1000 / 250000
[t-SNE] Computed conditional probabilities for sample 2000 / 250000
[t-SNE] Computed conditional probabilities for sample 3000 / 250000
[t-SNE] Computed conditional probabilities for sample 4000 / 250000
[t-SNE] Computed conditional probabilities for sample 5000 / 250000
[t-SNE] Computed conditional probabilities for sample 6000 / 250000
[t-SNE] Computed conditional probabilities for sample 7000 / 250000
[t-SNE] Computed conditional probabilities for sample 8000 / 250000
[t-SNE] Computed conditional probabilities for sample 9000 / 250000
[t-SNE] Computed conditional probabilities for sample 10000 / 250000
[t-SNE] Computed conditional probabilities for sample 11000 / 250000
[t-SNE] Computed conditional probabilities for sample 12000 / 250000
[t-SNE] Computed conditional probabilities for sample 13000 / 250000
[t-SNE] Computed conditional probabilities for sample 14000 / 250000
[t-SNE] Computed conditional probabilities for sample 15000 / 250000
[t-SNE] Computed conditional probabilities for sample 16000 / 250000
[t-SNE] Computed conditional probabilities for sample 17000 / 250000
[t-SNE] Computed conditional probabilities for sample 18000 / 250000
[t-SNE] Computed conditional probabilities for sample 19000 / 250000
[t-SNE] Computed conditional probabilities for sample 20000 / 250000
[t-SNE] Computed conditional probabilities for sample 21000 / 250000
[t-SNE] Computed conditional probabilities for sample 22000 / 250000
[t-SNE] Computed conditional probabilities for sample 23000 / 250000
[t-SNE] Computed conditional probabilities for sample 24000 / 250000
[t-SNE] Computed conditional probabilities for sample 25000 / 250000
[t-SNE] Computed conditional probabilities for sample 26000 / 250000
[t-SNE] Computed conditional probabilities for sample 27000 / 250000
[t-SNE] Computed conditional probabilities for sample 28000 / 250000
[t-SNE] Computed conditional probabilities for sample 29000 / 250000
[t-SNE] Computed conditional probabilities for sample 30000 / 250000
[t-SNE] Computed conditional probabilities for sample 31000 / 250000
[t-SNE] Computed conditional probabilities for sample 32000 / 250000
[t-SNE] Computed conditional probabilities for sample 33000 / 250000
[t-SNE] Computed conditional probabilities for sample 34000 / 250000
[t-SNE] Computed conditional probabilities for sample 35000 / 250000
[t-SNE] Computed conditional probabilities for sample 36000 / 250000
[t-SNE] Computed conditional probabilities for sample 37000 / 250000
[t-SNE] Computed conditional probabilities for sample 38000 / 250000
[t-SNE] Computed conditional probabilities for sample 39000 / 250000
[t-SNE] Computed conditional probabilities for sample 40000 / 250000
[t-SNE] Computed conditional probabilities for sample 41000 / 250000
[t-SNE] Computed conditional probabilities for sample 42000 / 250000
[t-SNE] Computed conditional probabilities for sample 43000 / 250000
[t-SNE] Computed conditional probabilities for sample 44000 / 250000
[t-SNE] Computed conditional probabilities for sample 45000 / 250000
[t-SNE] Computed conditional probabilities for sample 46000 / 250000
[t-SNE] Computed conditional probabilities for sample 47000 / 250000
[t-SNE] Computed conditional probabilities for sample 48000 / 250000
[t-SNE] Computed conditional probabilities for sample 49000 / 250000
[t-SNE] Computed conditional probabilities for sample 50000 / 250000
[t-SNE] Computed conditional probabilities for sample 51000 / 250000
[t-SNE] Computed conditional probabilities for sample 52000 / 250000
[t-SNE] Computed conditional probabilities for sample 53000 / 250000
[t-SNE] Computed conditional probabilities for sample 54000 / 250000
[t-SNE] Computed conditional probabilities for sample 55000 / 250000
[t-SNE] Computed conditional probabilities for sample 56000 / 250000
[t-SNE] Computed conditional probabilities for sample 57000 / 250000
[t-SNE] Computed conditional probabilities for sample 58000 / 250000
[t-SNE] Computed conditional probabilities for sample 59000 / 250000
[t-SNE] Computed conditional probabilities for sample 60000 / 250000
[t-SNE] Computed conditional probabilities for sample 61000 / 250000
[t-SNE] Computed conditional probabilities for sample 62000 / 250000
[t-SNE] Computed conditional probabilities for sample 63000 / 250000
[t-SNE] Computed conditional probabilities for sample 64000 / 250000
[t-SNE] Computed conditional probabilities for sample 65000 / 250000
[t-SNE] Computed conditional probabilities for sample 66000 / 250000
[t-SNE] Computed conditional probabilities for sample 67000 / 250000
[t-SNE] Computed conditional probabilities for sample 68000 / 250000
[t-SNE] Computed conditional probabilities for sample 69000 / 250000
[t-SNE] Computed conditional probabilities for sample 70000 / 250000
[t-SNE] Computed conditional probabilities for sample 71000 / 250000
[t-SNE] Computed conditional probabilities for sample 72000 / 250000
[t-SNE] Computed conditional probabilities for sample 73000 / 250000
[t-SNE] Computed conditional probabilities for sample 74000 / 250000
[t-SNE] Computed conditional probabilities for sample 75000 / 250000
[t-SNE] Computed conditional probabilities for sample 76000 / 250000
[t-SNE] Computed conditional probabilities for sample 77000 / 250000
[t-SNE] Computed conditional probabilities for sample 78000 / 250000
[t-SNE] Computed conditional probabilities for sample 79000 / 250000
[t-SNE] Computed conditional probabilities for sample 80000 / 250000
[t-SNE] Computed conditional probabilities for sample 81000 / 250000
[t-SNE] Computed conditional probabilities for sample 82000 / 250000
[t-SNE] Computed conditional probabilities for sample 83000 / 250000
[t-SNE] Computed conditional probabilities for sample 84000 / 250000
[t-SNE] Computed conditional probabilities for sample 85000 / 250000
[t-SNE] Computed conditional probabilities for sample 86000 / 250000
[t-SNE] Computed conditional probabilities for sample 87000 / 250000
[t-SNE] Computed conditional probabilities for sample 88000 / 250000
[t-SNE] Computed conditional probabilities for sample 89000 / 250000
[t-SNE] Computed conditional probabilities for sample 90000 / 250000
[t-SNE] Computed conditional probabilities for sample 91000 / 250000
[t-SNE] Computed conditional probabilities for sample 92000 / 250000
[t-SNE] Computed conditional probabilities for sample 93000 / 250000
[t-SNE] Computed conditional probabilities for sample 94000 / 250000
[t-SNE] Computed conditional probabilities for sample 95000 / 250000
[t-SNE] Computed conditional probabilities for sample 96000 / 250000
[t-SNE] Computed conditional probabilities for sample 97000 / 250000
[t-SNE] Computed conditional probabilities for sample 98000 / 250000
[t-SNE] Computed conditional probabilities for sample 99000 / 250000
[t-SNE] Computed conditional probabilities for sample 100000 / 250000
[t-SNE] Computed conditional probabilities for sample 101000 / 250000
[t-SNE] Computed conditional probabilities for sample 102000 / 250000
[t-SNE] Computed conditional probabilities for sample 103000 / 250000
[t-SNE] Computed conditional probabilities for sample 104000 / 250000
[t-SNE] Computed conditional probabilities for sample 105000 / 250000
[t-SNE] Computed conditional probabilities for sample 106000 / 250000
[t-SNE] Computed conditional probabilities for sample 107000 / 250000
[t-SNE] Computed conditional probabilities for sample 108000 / 250000
[t-SNE] Computed conditional probabilities for sample 109000 / 250000
[t-SNE] Computed conditional probabilities for sample 110000 / 250000
[t-SNE] Computed conditional probabilities for sample 111000 / 250000
[t-SNE] Computed conditional probabilities for sample 112000 / 250000
[t-SNE] Computed conditional probabilities for sample 113000 / 250000
[t-SNE] Computed conditional probabilities for sample 114000 / 250000
[t-SNE] Computed conditional probabilities for sample 115000 / 250000
[t-SNE] Computed conditional probabilities for sample 116000 / 250000
[t-SNE] Computed conditional probabilities for sample 117000 / 250000
[t-SNE] Computed conditional probabilities for sample 118000 / 250000
[t-SNE] Computed conditional probabilities for sample 119000 / 250000
[t-SNE] Computed conditional probabilities for sample 120000 / 250000
[t-SNE] Computed conditional probabilities for sample 121000 / 250000
[t-SNE] Computed conditional probabilities for sample 122000 / 250000
[t-SNE] Computed conditional probabilities for sample 123000 / 250000
[t-SNE] Computed conditional probabilities for sample 124000 / 250000
[t-SNE] Computed conditional probabilities for sample 125000 / 250000
[t-SNE] Computed conditional probabilities for sample 126000 / 250000
[t-SNE] Computed conditional probabilities for sample 127000 / 250000
[t-SNE] Computed conditional probabilities for sample 128000 / 250000
[t-SNE] Computed conditional probabilities for sample 129000 / 250000
[t-SNE] Computed conditional probabilities for sample 130000 / 250000
[t-SNE] Computed conditional probabilities for sample 131000 / 250000
[t-SNE] Computed conditional probabilities for sample 132000 / 250000
[t-SNE] Computed conditional probabilities for sample 133000 / 250000

[t-SNE] Computed conditional probabilities for sample 134000 / 250000
[t-SNE] Computed conditional probabilities for sample 135000 / 250000
[t-SNE] Computed conditional probabilities for sample 136000 / 250000
[t-SNE] Computed conditional probabilities for sample 137000 / 250000
[t-SNE] Computed conditional probabilities for sample 138000 / 250000
[t-SNE] Computed conditional probabilities for sample 139000 / 250000
[t-SNE] Computed conditional probabilities for sample 140000 / 250000
[t-SNE] Computed conditional probabilities for sample 141000 / 250000
[t-SNE] Computed conditional probabilities for sample 142000 / 250000
[t-SNE] Computed conditional probabilities for sample 143000 / 250000
[t-SNE] Computed conditional probabilities for sample 144000 / 250000
[t-SNE] Computed conditional probabilities for sample 145000 / 250000
[t-SNE] Computed conditional probabilities for sample 146000 / 250000
[t-SNE] Computed conditional probabilities for sample 147000 / 250000
[t-SNE] Computed conditional probabilities for sample 148000 / 250000
[t-SNE] Computed conditional probabilities for sample 149000 / 250000
[t-SNE] Computed conditional probabilities for sample 150000 / 250000
[t-SNE] Computed conditional probabilities for sample 151000 / 250000
[t-SNE] Computed conditional probabilities for sample 152000 / 250000
[t-SNE] Computed conditional probabilities for sample 153000 / 250000
[t-SNE] Computed conditional probabilities for sample 154000 / 250000
[t-SNE] Computed conditional probabilities for sample 155000 / 250000
[t-SNE] Computed conditional probabilities for sample 156000 / 250000
[t-SNE] Computed conditional probabilities for sample 157000 / 250000
[t-SNE] Computed conditional probabilities for sample 158000 / 250000
[t-SNE] Computed conditional probabilities for sample 159000 / 250000
[t-SNE] Computed conditional probabilities for sample 160000 / 250000
[t-SNE] Computed conditional probabilities for sample 161000 / 250000
[t-SNE] Computed conditional probabilities for sample 162000 / 250000
[t-SNE] Computed conditional probabilities for sample 163000 / 250000
[t-SNE] Computed conditional probabilities for sample 164000 / 250000
[t-SNE] Computed conditional probabilities for sample 165000 / 250000
[t-SNE] Computed conditional probabilities for sample 166000 / 250000
[t-SNE] Computed conditional probabilities for sample 167000 / 250000
[t-SNE] Computed conditional probabilities for sample 168000 / 250000
[t-SNE] Computed conditional probabilities for sample 169000 / 250000
[t-SNE] Computed conditional probabilities for sample 170000 / 250000
[t-SNE] Computed conditional probabilities for sample 171000 / 250000
[t-SNE] Computed conditional probabilities for sample 172000 / 250000
[t-SNE] Computed conditional probabilities for sample 173000 / 250000
[t-SNE] Computed conditional probabilities for sample 174000 / 250000
[t-SNE] Computed conditional probabilities for sample 175000 / 250000
[t-SNE] Computed conditional probabilities for sample 176000 / 250000
[t-SNE] Computed conditional probabilities for sample 177000 / 250000
[t-SNE] Computed conditional probabilities for sample 178000 / 250000
[t-SNE] Computed conditional probabilities for sample 179000 / 250000
[t-SNE] Computed conditional probabilities for sample 180000 / 250000
[t-SNE] Computed conditional probabilities for sample 181000 / 250000
[t-SNE] Computed conditional probabilities for sample 182000 / 250000
[t-SNE] Computed conditional probabilities for sample 183000 / 250000
[t-SNE] Computed conditional probabilities for sample 184000 / 250000
[t-SNE] Computed conditional probabilities for sample 185000 / 250000
[t-SNE] Computed conditional probabilities for sample 186000 / 250000
[t-SNE] Computed conditional probabilities for sample 187000 / 250000
[t-SNE] Computed conditional probabilities for sample 188000 / 250000
[t-SNE] Computed conditional probabilities for sample 189000 / 250000
[t-SNE] Computed conditional probabilities for sample 190000 / 250000
[t-SNE] Computed conditional probabilities for sample 191000 / 250000
[t-SNE] Computed conditional probabilities for sample 192000 / 250000
[t-SNE] Computed conditional probabilities for sample 193000 / 250000
[t-SNE] Computed conditional probabilities for sample 194000 / 250000
[t-SNE] Computed conditional probabilities for sample 195000 / 250000
[t-SNE] Computed conditional probabilities for sample 196000 / 250000
[t-SNE] Computed conditional probabilities for sample 197000 / 250000
[t-SNE] Computed conditional probabilities for sample 198000 / 250000
[t-SNE] Computed conditional probabilities for sample 199000 / 250000
[t-SNE] Computed conditional probabilities for sample 200000 / 250000
[t-SNE] Computed conditional probabilities for sample 201000 / 250000
[t-SNE] Computed conditional probabilities for sample 202000 / 250000
[t-SNE] Computed conditional probabilities for sample 203000 / 250000
[t-SNE] Computed conditional probabilities for sample 204000 / 250000
[t-SNE] Computed conditional probabilities for sample 205000 / 250000
[t-SNE] Computed conditional probabilities for sample 206000 / 250000
[t-SNE] Computed conditional probabilities for sample 207000 / 250000
[t-SNE] Computed conditional probabilities for sample 208000 / 250000
[t-SNE] Computed conditional probabilities for sample 209000 / 250000
[t-SNE] Computed conditional probabilities for sample 210000 / 250000
[t-SNE] Computed conditional probabilities for sample 211000 / 250000
[t-SNE] Computed conditional probabilities for sample 212000 / 250000
[t-SNE] Computed conditional probabilities for sample 213000 / 250000
[t-SNE] Computed conditional probabilities for sample 214000 / 250000
[t-SNE] Computed conditional probabilities for sample 215000 / 250000
[t-SNE] Computed conditional probabilities for sample 216000 / 250000
[t-SNE] Computed conditional probabilities for sample 217000 / 250000
[t-SNE] Computed conditional probabilities for sample 218000 / 250000
[t-SNE] Computed conditional probabilities for sample 219000 / 250000
[t-SNE] Computed conditional probabilities for sample 220000 / 250000
[t-SNE] Computed conditional probabilities for sample 221000 / 250000
[t-SNE] Computed conditional probabilities for sample 222000 / 250000
[t-SNE] Computed conditional probabilities for sample 223000 / 250000
[t-SNE] Computed conditional probabilities for sample 224000 / 250000
[t-SNE] Computed conditional probabilities for sample 225000 / 250000
[t-SNE] Computed conditional probabilities for sample 226000 / 250000
[t-SNE] Computed conditional probabilities for sample 227000 / 250000
[t-SNE] Computed conditional probabilities for sample 228000 / 250000
[t-SNE] Computed conditional probabilities for sample 229000 / 250000
[t-SNE] Computed conditional probabilities for sample 230000 / 250000
[t-SNE] Computed conditional probabilities for sample 231000 / 250000
[t-SNE] Computed conditional probabilities for sample 232000 / 250000
[t-SNE] Computed conditional probabilities for sample 233000 / 250000
[t-SNE] Computed conditional probabilities for sample 234000 / 250000
[t-SNE] Computed conditional probabilities for sample 235000 / 250000
[t-SNE] Computed conditional probabilities for sample 236000 / 250000
[t-SNE] Computed conditional probabilities for sample 237000 / 250000
[t-SNE] Computed conditional probabilities for sample 238000 / 250000
[t-SNE] Computed conditional probabilities for sample 239000 / 250000
[t-SNE] Computed conditional probabilities for sample 240000 / 250000
[t-SNE] Computed conditional probabilities for sample 241000 / 250000
[t-SNE] Computed conditional probabilities for sample 242000 / 250000
[t-SNE] Computed conditional probabilities for sample 243000 / 250000
[t-SNE] Computed conditional probabilities for sample 244000 / 250000
[t-SNE] Computed conditional probabilities for sample 245000 / 250000
[t-SNE] Computed conditional probabilities for sample 246000 / 250000
[t-SNE] Computed conditional probabilities for sample 247000 / 250000
[t-SNE] Computed conditional probabilities for sample 248000 / 250000
[t-SNE] Computed conditional probabilities for sample 249000 / 250000
[t-SNE] Computed conditional probabilities for sample 250000 / 250000
[t-SNE] Mean sigma: 0.387824

[t-SNE] KL divergence after 50 iterations with early exaggeration: 125.084518
[t-SNE] KL divergence after 250 iterations: 6.187023
t-SNE done! Time elapsed: 1069.5455431938171 seconds

tsne_df1 = pd.DataFrame()

tsne_df1.to_csv('tsne1.csv', index=False)

​

​

​

​

UMAP

#creating a new dataframe with the robust scaled values

scaled_df = pd.DataFrame(scaled_data, columns=['DER_mass_MMC', 'DER_mass_transverse_met_lep',

scaled_df = pd.concat([scaled_df, df[['Weight', 'Label']]], axis=1)

import umap.umap_ as umap

reducer = umap.UMAP()

#applying UMAP on scaled data

/home/013729625/dsenv/lib64/python3.6/site-packages/numba/typed_passes.py:271: NumbaPerformanceWarning:


The keyword argument 'parallel=True' was specified but no transformation for parallel execution was possible.

To find out why, try turning on parallel diagnostics, see http://numba.pydata.org/numba-doc/latest/user/parallel.html#diagnostics for help.

File "dsenv/lib64/python3.6/site-packages/umap/rp_tree.py", line 135:
@numba.njit(fastmath=True, nogil=True, parallel=True)
def euclidean_random_projection_split(data, indices, rng_state):
^


/home/013729625/dsenv/lib64/python3.6/site-packages/numba/typed_passes.py:271: NumbaPerformanceWarning:


The keyword argument 'parallel=True' was specified but no transformation for parallel execution was possible.

To find out why, try turning on parallel diagnostics, see http://numba.pydata.org/numba-doc/latest/user/parallel.html#diagnostics for help.

File "dsenv/lib64/python3.6/site-packages/umap/nndescent.py", line 47:
    @numba.njit(parallel=True)
    def nn_descent(
    ^

(250000, 2)

x

umap_df = pd.DataFrame()

umap_df.to_csv('umap_df.csv', index=False)

embedding_pca = reducer.fit_transform(pca_data)

/home/013729625/dsenv/lib64/python3.6/site-packages/numba/typed_passes.py:271: NumbaPerformanceWarning:


The keyword argument 'parallel=True' was specified but no transformation for parallel execution was possible.

To find out why, try turning on parallel diagnostics, see http://numba.pydata.org/numba-doc/latest/user/parallel.html#diagnostics for help.

File "dsenv/lib64/python3.6/site-packages/umap/nndescent.py", line 47:
    @numba.njit(parallel=True)
    def nn_descent(
    ^

embedding_pca.shape

(250000, 2)

plt.figure(figsize=(12,8))

​

std_data = scaler.fit_transform(df.drop(['EventId', 'Label'], axis=1))

std_df = pd.DataFrame(std_data, columns=['DER_mass_MMC', 'DER_mass_transverse_met_lep',

import time

#testing t-SNE on first 10000 rows of standard scaled data

time_start = time.time()

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 10000 samples in 0.023s...
[t-SNE] Computed neighbors for 10000 samples in 3.916s...
[t-SNE] Computed conditional probabilities for sample 1000 / 10000
[t-SNE] Computed conditional probabilities for sample 2000 / 10000
[t-SNE] Computed conditional probabilities for sample 3000 / 10000
[t-SNE] Computed conditional probabilities for sample 4000 / 10000
[t-SNE] Computed conditional probabilities for sample 5000 / 10000
[t-SNE] Computed conditional probabilities for sample 6000 / 10000
[t-SNE] Computed conditional probabilities for sample 7000 / 10000
[t-SNE] Computed conditional probabilities for sample 8000 / 10000
[t-SNE] Computed conditional probabilities for sample 9000 / 10000
[t-SNE] Computed conditional probabilities for sample 10000 / 10000
[t-SNE] Mean sigma: 0.818923
[t-SNE] KL divergence after 250 iterations with early exaggeration: 80.969681
[t-SNE] KL divergence after 251 iterations: 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.000000
t-SNE done! Time elapsed: 102.87509083747864 seconds

tsne_arr2

array([[ 1.1844845 ,  1.2650347 , -1.3530798 ],
       [ 1.0320665 , -1.0249159 ,  1.1623231 ],
       [-1.8379277 , -1.7901961 , -2.386018  ],
       ...,
       [ 1.4897859 ,  0.77739704, -1.1057904 ],
       [-1.8114654 ,  0.6289672 ,  1.5614039 ],
       [-1.8921157 ,  0.57281023,  1.5513233 ]], dtype=float32)

x

plt.figure(figsize=(14,10))

<matplotlib.axes._subplots.AxesSubplot at 0x7f773875dd30>

​

​

​

​

xxxxxxxxxx

df.head()

x

(array([1.77457e+05, 0.00000e+00, 0.00000e+00, 0.00000e+00, 0.00000e+00,
        0.00000e+00, 0.00000e+00, 0.00000e+00, 7.95200e+03, 2.44800e+04,
        1.23040e+04, 7.50000e+03, 5.26300e+03, 3.69900e+03, 2.65800e+03,
        2.08000e+03, 1.53300e+03, 1.16600e+03, 8.66000e+02, 7.09000e+02,
        5.16000e+02, 3.96000e+02, 3.59000e+02, 2.52000e+02, 1.92000e+02,
        1.62000e+02, 1.04000e+02, 8.30000e+01, 6.50000e+01, 5.00000e+01,
        3.40000e+01, 3.70000e+01, 1.80000e+01, 1.90000e+01, 1.30000e+01,
        6.00000e+00, 6.00000e+00, 6.00000e+00, 4.00000e+00, 4.00000e+00,
        1.00000e+00, 2.00000e+00, 1.00000e+00, 1.00000e+00, 0.00000e+00,
        0.00000e+00, 1.00000e+00, 0.00000e+00, 0.00000e+00, 1.00000e+00]),
 array([-999.     , -879.52042, -760.04084, -640.56126, -521.08168,
        -401.6021 , -282.12252, -162.64294,  -43.16336,   76.31622,
         195.7958 ,  315.27538,  434.75496,  554.23454,  673.71412,
         793.1937 ,  912.67328, 1032.15286, 1151.63244, 1271.11202,
        1390.5916 , 1510.07118, 1629.55076, 1749.03034, 1868.50992,
        1987.9895 , 2107.46908, 2226.94866, 2346.42824, 2465.90782,
        2585.3874 , 2704.86698, 2824.34656, 2943.82614, 3063.30572,
        3182.7853 , 3302.26488, 3421.74446, 3541.22404, 3660.70362,
        3780.1832 , 3899.66278, 4019.14236, 4138.62194, 4258.10152,
        4377.5811 , 4497.06068, 4616.54026, 4736.01984, 4855.49942,
        4974.979  ]),
 <a list of 50 Patch objects>)

xxxxxxxxxx

df_jet_0 = df.copy()

#grouping the training data into 3 groups using the PRI_jet_num feature as filter

x

#separting data with PRI_jet_num = 1

#separting data with PRI_jet_num 2 and PRI_jet_num = 3

df_jet_0.columns

Index(['EventId', 'DER_mass_MMC', 'DER_mass_transverse_met_lep',
       'DER_mass_vis', 'DER_pt_h', 'DER_deltaeta_jet_jet', 'DER_mass_jet_jet',
       'DER_prodeta_jet_jet', 'DER_deltar_tau_lep', 'DER_pt_tot', 'DER_sum_pt',
       'DER_pt_ratio_lep_tau', 'DER_met_phi_centrality',
       'DER_lep_eta_centrality', 'PRI_tau_pt', 'PRI_tau_eta', 'PRI_tau_phi',
       'PRI_lep_pt', 'PRI_lep_eta', 'PRI_lep_phi', 'PRI_met', 'PRI_met_phi',
       'PRI_met_sumet', 'PRI_jet_num', 'PRI_jet_leading_pt',
       'PRI_jet_leading_eta', 'PRI_jet_leading_phi', 'PRI_jet_subleading_pt',
       'PRI_jet_subleading_eta', 'PRI_jet_subleading_phi', 'PRI_jet_all_pt',
       'Weight', 'Label'],
      dtype='object')

#for data with PRI_jet_num = 0, the following columns have all null values

#for data with PRI_jet_num = 1, the following columns have all null values

#PRI_jet_num only has a constant value

from sklearn.preprocessing import StandardScaler

s_scaler = StandardScaler()

x

pca_2 = PCA(n_components=2)

#applying PCA on grouped data

[0.18515646 0.16167186]
[0.24742629 0.10978555]
[0.20681785 0.11292406]

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f7603302550>

plt.figure(figsize=(14,10))

<matplotlib.axes._subplots.AxesSubplot at 0x7f76047074e0>

plt.figure(figsize=(14,10))

<matplotlib.axes._subplots.AxesSubplot at 0x7f7604611cf8>

#applying PCA with 3 components

pca_3 = PCA(n_components=3)

x

[0.18515646 0.16167186 0.13806351]
[0.24742629 0.10978554 0.08655618]
[0.20681785 0.11292405 0.07750242]

from sklearn.preprocessing import RobustScaler

r_scaler = RobustScaler()

#using Robust Scaler to check performance of PCA

x

[0.53405051 0.14226634]
[0.75516214 0.10120474]
[0.68082688 0.08636026]

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f7604727cf8>

plt.figure(figsize=(14,10))

<matplotlib.axes._subplots.AxesSubplot at 0x7f7604738710>

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f76049a4198>

rpca3_jet_0 = pca_3.fit_transform(rscaled_jet_0)

[0.53405051 0.14226634 0.07950811]
[0.75516214 0.10120474 0.03680352]
[0.68082688 0.08636026 0.04351135]

​

pca_10 = PCA(n_components=10)

#with 10 components PCA covers around 96% of the total variance

0.9676285593445852
0.9706432481751432
0.9353877928385876

​

​

xxxxxxxxxx

xxxxxxxxxx

from sklearn.model_selection import train_test_split

df_jet_0.columns

Index(['EventId', 'DER_mass_MMC', 'DER_mass_transverse_met_lep',
       'DER_mass_vis', 'DER_pt_h', 'DER_deltar_tau_lep', 'DER_pt_tot',
       'DER_sum_pt', 'DER_pt_ratio_lep_tau', 'DER_met_phi_centrality',
       'PRI_tau_pt', 'PRI_tau_eta', 'PRI_tau_phi', 'PRI_lep_pt', 'PRI_lep_eta',
       'PRI_lep_phi', 'PRI_met', 'PRI_met_phi', 'PRI_met_sumet',
       'PRI_jet_all_pt', 'Weight', 'Label'],
      dtype='object')

x

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=101)

from sklearn.linear_model import LogisticRegression

log_model = LogisticRegression()

log_model.fit(X_train, y_train)

/home/013729625/dsenv/lib64/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning:

Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.

LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False)

predictions = log_model.predict(X_test)

from sklearn.metrics import classification_report, confusion_matrix

#Performance metrics for data PRI_jet_num = 0

[[27344  2458]
 [ 4473  5691]]
              precision    recall  f1-score   support

           b       0.86      0.92      0.89     29802
           s       0.70      0.56      0.62     10164

    accuracy                           0.83     39966
   macro avg       0.78      0.74      0.75     39966
weighted avg       0.82      0.83      0.82     39966

#Applying logistic regression on data with PRI_jet_num = 1

/home/013729625/dsenv/lib64/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning:

Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.

[[16647  3301]
 [ 5656  5414]]
              precision    recall  f1-score   support

           b       0.75      0.83      0.79     19948
           s       0.62      0.49      0.55     11070

    accuracy                           0.71     31018
   macro avg       0.68      0.66      0.67     31018
weighted avg       0.70      0.71      0.70     31018

#Applying logistic regression on data with PRI_jet_num > 1

/home/013729625/dsenv/lib64/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning:

Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.

[[12808  3260]
 [ 4293  8657]]
              precision    recall  f1-score   support

           b       0.75      0.80      0.77     16068
           s       0.73      0.67      0.70     12950

    accuracy                           0.74     29018
   macro avg       0.74      0.73      0.73     29018
weighted avg       0.74      0.74      0.74     29018

​

xxxxxxxxxx

​

from sklearn.ensemble import RandomForestClassifier

rfc = RandomForestClassifier(n_estimators=300)

x

[[27786  2016]
 [ 3947  6217]]
              precision    recall  f1-score   support

           b       0.88      0.93      0.90     29802
           s       0.76      0.61      0.68     10164

    accuracy                           0.85     39966
   macro avg       0.82      0.77      0.79     39966
weighted avg       0.84      0.85      0.85     39966

#Applying random forest classifier on data with PRI_jet_num = 1

[[17618  2330]
 [ 3456  7614]]
              precision    recall  f1-score   support

           b       0.84      0.88      0.86     19948
           s       0.77      0.69      0.72     11070

    accuracy                           0.81     31018
   macro avg       0.80      0.79      0.79     31018
weighted avg       0.81      0.81      0.81     31018

#Applying random forest classifier on data with PRI_jet_num > 1

[[14179  1889]
 [ 2408 10542]]
              precision    recall  f1-score   support

           b       0.85      0.88      0.87     16068
           s       0.85      0.81      0.83     12950

    accuracy                           0.85     29018
   macro avg       0.85      0.85      0.85     29018
weighted avg       0.85      0.85      0.85     29018

​

xxxxxxxxxx

​

from xgboost import XGBClassifier

xgb = XGBClassifier(n_estimators=300)

x

[[27627  2175]
 [ 3761  6403]]
              precision    recall  f1-score   support

           b       0.88      0.93      0.90     29802
           s       0.75      0.63      0.68     10164

    accuracy                           0.85     39966
   macro avg       0.81      0.78      0.79     39966
weighted avg       0.85      0.85      0.85     39966

#Applying XGBoost classifier on data with PRI_jet_num = 1

[[17533  2415]
 [ 3210  7860]]
              precision    recall  f1-score   support

           b       0.85      0.88      0.86     19948
           s       0.76      0.71      0.74     11070

    accuracy                           0.82     31018
   macro avg       0.81      0.79      0.80     31018
weighted avg       0.82      0.82      0.82     31018

#Applying XGBoost classifier on data with PRI_jet_num > 1

[[14002  2066]
 [ 2238 10712]]
              precision    recall  f1-score   support

           b       0.86      0.87      0.87     16068
           s       0.84      0.83      0.83     12950

    accuracy                           0.85     29018
   macro avg       0.85      0.85      0.85     29018
weighted avg       0.85      0.85      0.85     29018

​

xxxxxxxxxx

​

x

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 99913 samples in 0.425s...
[t-SNE] Computed neighbors for 99913 samples in 21.384s...
[t-SNE] Computed conditional probabilities for sample 1000 / 99913
[t-SNE] Computed conditional probabilities for sample 2000 / 99913
[t-SNE] Computed conditional probabilities for sample 3000 / 99913
[t-SNE] Computed conditional probabilities for sample 4000 / 99913
[t-SNE] Computed conditional probabilities for sample 5000 / 99913
[t-SNE] Computed conditional probabilities for sample 6000 / 99913
[t-SNE] Computed conditional probabilities for sample 7000 / 99913
[t-SNE] Computed conditional probabilities for sample 8000 / 99913
[t-SNE] Computed conditional probabilities for sample 9000 / 99913
[t-SNE] Computed conditional probabilities for sample 10000 / 99913
[t-SNE] Computed conditional probabilities for sample 11000 / 99913
[t-SNE] Computed conditional probabilities for sample 12000 / 99913
[t-SNE] Computed conditional probabilities for sample 13000 / 99913
[t-SNE] Computed conditional probabilities for sample 14000 / 99913
[t-SNE] Computed conditional probabilities for sample 15000 / 99913
[t-SNE] Computed conditional probabilities for sample 16000 / 99913
[t-SNE] Computed conditional probabilities for sample 17000 / 99913
[t-SNE] Computed conditional probabilities for sample 18000 / 99913
[t-SNE] Computed conditional probabilities for sample 19000 / 99913
[t-SNE] Computed conditional probabilities for sample 20000 / 99913
[t-SNE] Computed conditional probabilities for sample 21000 / 99913
[t-SNE] Computed conditional probabilities for sample 22000 / 99913
[t-SNE] Computed conditional probabilities for sample 23000 / 99913
[t-SNE] Computed conditional probabilities for sample 24000 / 99913
[t-SNE] Computed conditional probabilities for sample 25000 / 99913
[t-SNE] Computed conditional probabilities for sample 26000 / 99913
[t-SNE] Computed conditional probabilities for sample 27000 / 99913
[t-SNE] Computed conditional probabilities for sample 28000 / 99913
[t-SNE] Computed conditional probabilities for sample 29000 / 99913
[t-SNE] Computed conditional probabilities for sample 30000 / 99913
[t-SNE] Computed conditional probabilities for sample 31000 / 99913
[t-SNE] Computed conditional probabilities for sample 32000 / 99913
[t-SNE] Computed conditional probabilities for sample 33000 / 99913
[t-SNE] Computed conditional probabilities for sample 34000 / 99913
[t-SNE] Computed conditional probabilities for sample 35000 / 99913
[t-SNE] Computed conditional probabilities for sample 36000 / 99913
[t-SNE] Computed conditional probabilities for sample 37000 / 99913
[t-SNE] Computed conditional probabilities for sample 38000 / 99913
[t-SNE] Computed conditional probabilities for sample 39000 / 99913
[t-SNE] Computed conditional probabilities for sample 40000 / 99913
[t-SNE] Computed conditional probabilities for sample 41000 / 99913
[t-SNE] Computed conditional probabilities for sample 42000 / 99913
[t-SNE] Computed conditional probabilities for sample 43000 / 99913
[t-SNE] Computed conditional probabilities for sample 44000 / 99913
[t-SNE] Computed conditional probabilities for sample 45000 / 99913
[t-SNE] Computed conditional probabilities for sample 46000 / 99913
[t-SNE] Computed conditional probabilities for sample 47000 / 99913
[t-SNE] Computed conditional probabilities for sample 48000 / 99913
[t-SNE] Computed conditional probabilities for sample 49000 / 99913
[t-SNE] Computed conditional probabilities for sample 50000 / 99913
[t-SNE] Computed conditional probabilities for sample 51000 / 99913
[t-SNE] Computed conditional probabilities for sample 52000 / 99913
[t-SNE] Computed conditional probabilities for sample 53000 / 99913
[t-SNE] Computed conditional probabilities for sample 54000 / 99913
[t-SNE] Computed conditional probabilities for sample 55000 / 99913
[t-SNE] Computed conditional probabilities for sample 56000 / 99913
[t-SNE] Computed conditional probabilities for sample 57000 / 99913
[t-SNE] Computed conditional probabilities for sample 58000 / 99913
[t-SNE] Computed conditional probabilities for sample 59000 / 99913
[t-SNE] Computed conditional probabilities for sample 60000 / 99913
[t-SNE] Computed conditional probabilities for sample 61000 / 99913
[t-SNE] Computed conditional probabilities for sample 62000 / 99913
[t-SNE] Computed conditional probabilities for sample 63000 / 99913
[t-SNE] Computed conditional probabilities for sample 64000 / 99913
[t-SNE] Computed conditional probabilities for sample 65000 / 99913
[t-SNE] Computed conditional probabilities for sample 66000 / 99913
[t-SNE] Computed conditional probabilities for sample 67000 / 99913
[t-SNE] Computed conditional probabilities for sample 68000 / 99913
[t-SNE] Computed conditional probabilities for sample 69000 / 99913
[t-SNE] Computed conditional probabilities for sample 70000 / 99913
[t-SNE] Computed conditional probabilities for sample 71000 / 99913
[t-SNE] Computed conditional probabilities for sample 72000 / 99913
[t-SNE] Computed conditional probabilities for sample 73000 / 99913
[t-SNE] Computed conditional probabilities for sample 74000 / 99913
[t-SNE] Computed conditional probabilities for sample 75000 / 99913
[t-SNE] Computed conditional probabilities for sample 76000 / 99913
[t-SNE] Computed conditional probabilities for sample 77000 / 99913
[t-SNE] Computed conditional probabilities for sample 78000 / 99913
[t-SNE] Computed conditional probabilities for sample 79000 / 99913
[t-SNE] Computed conditional probabilities for sample 80000 / 99913
[t-SNE] Computed conditional probabilities for sample 81000 / 99913
[t-SNE] Computed conditional probabilities for sample 82000 / 99913
[t-SNE] Computed conditional probabilities for sample 83000 / 99913
[t-SNE] Computed conditional probabilities for sample 84000 / 99913
[t-SNE] Computed conditional probabilities for sample 85000 / 99913
[t-SNE] Computed conditional probabilities for sample 86000 / 99913
[t-SNE] Computed conditional probabilities for sample 87000 / 99913
[t-SNE] Computed conditional probabilities for sample 88000 / 99913
[t-SNE] Computed conditional probabilities for sample 89000 / 99913
[t-SNE] Computed conditional probabilities for sample 90000 / 99913
[t-SNE] Computed conditional probabilities for sample 91000 / 99913
[t-SNE] Computed conditional probabilities for sample 92000 / 99913
[t-SNE] Computed conditional probabilities for sample 93000 / 99913
[t-SNE] Computed conditional probabilities for sample 94000 / 99913
[t-SNE] Computed conditional probabilities for sample 95000 / 99913
[t-SNE] Computed conditional probabilities for sample 96000 / 99913
[t-SNE] Computed conditional probabilities for sample 97000 / 99913
[t-SNE] Computed conditional probabilities for sample 98000 / 99913
[t-SNE] Computed conditional probabilities for sample 99000 / 99913
[t-SNE] Computed conditional probabilities for sample 99913 / 99913
[t-SNE] Mean sigma: 6.305854
[t-SNE] KL divergence after 250 iterations with early exaggeration: 100.063934
[t-SNE] KL divergence after 251 iterations: 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.000000
t-SNE done! Time elapsed: 475.10227131843567 seconds

df_jet_0.to_csv('df_jet_0.csv', index=False)

plt.figure(figsize=(14,10))

<matplotlib.axes._subplots.AxesSubplot at 0x7f7604cb5668>

#Applying t-SNE on data with PRI_jet_num = 1

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 77544 samples in 0.145s...
[t-SNE] Computed neighbors for 77544 samples in 26.416s...
[t-SNE] Computed conditional probabilities for sample 1000 / 77544
[t-SNE] Computed conditional probabilities for sample 2000 / 77544
[t-SNE] Computed conditional probabilities for sample 3000 / 77544
[t-SNE] Computed conditional probabilities for sample 4000 / 77544
[t-SNE] Computed conditional probabilities for sample 5000 / 77544
[t-SNE] Computed conditional probabilities for sample 6000 / 77544
[t-SNE] Computed conditional probabilities for sample 7000 / 77544
[t-SNE] Computed conditional probabilities for sample 8000 / 77544
[t-SNE] Computed conditional probabilities for sample 9000 / 77544
[t-SNE] Computed conditional probabilities for sample 10000 / 77544
[t-SNE] Computed conditional probabilities for sample 11000 / 77544
[t-SNE] Computed conditional probabilities for sample 12000 / 77544
[t-SNE] Computed conditional probabilities for sample 13000 / 77544
[t-SNE] Computed conditional probabilities for sample 14000 / 77544
[t-SNE] Computed conditional probabilities for sample 15000 / 77544
[t-SNE] Computed conditional probabilities for sample 16000 / 77544
[t-SNE] Computed conditional probabilities for sample 17000 / 77544
[t-SNE] Computed conditional probabilities for sample 18000 / 77544
[t-SNE] Computed conditional probabilities for sample 19000 / 77544
[t-SNE] Computed conditional probabilities for sample 20000 / 77544
[t-SNE] Computed conditional probabilities for sample 21000 / 77544
[t-SNE] Computed conditional probabilities for sample 22000 / 77544
[t-SNE] Computed conditional probabilities for sample 23000 / 77544
[t-SNE] Computed conditional probabilities for sample 24000 / 77544
[t-SNE] Computed conditional probabilities for sample 25000 / 77544
[t-SNE] Computed conditional probabilities for sample 26000 / 77544
[t-SNE] Computed conditional probabilities for sample 27000 / 77544
[t-SNE] Computed conditional probabilities for sample 28000 / 77544
[t-SNE] Computed conditional probabilities for sample 29000 / 77544
[t-SNE] Computed conditional probabilities for sample 30000 / 77544
[t-SNE] Computed conditional probabilities for sample 31000 / 77544
[t-SNE] Computed conditional probabilities for sample 32000 / 77544
[t-SNE] Computed conditional probabilities for sample 33000 / 77544
[t-SNE] Computed conditional probabilities for sample 34000 / 77544
[t-SNE] Computed conditional probabilities for sample 35000 / 77544
[t-SNE] Computed conditional probabilities for sample 36000 / 77544
[t-SNE] Computed conditional probabilities for sample 37000 / 77544
[t-SNE] Computed conditional probabilities for sample 38000 / 77544
[t-SNE] Computed conditional probabilities for sample 39000 / 77544
[t-SNE] Computed conditional probabilities for sample 40000 / 77544
[t-SNE] Computed conditional probabilities for sample 41000 / 77544
[t-SNE] Computed conditional probabilities for sample 42000 / 77544
[t-SNE] Computed conditional probabilities for sample 43000 / 77544
[t-SNE] Computed conditional probabilities for sample 44000 / 77544
[t-SNE] Computed conditional probabilities for sample 45000 / 77544
[t-SNE] Computed conditional probabilities for sample 46000 / 77544
[t-SNE] Computed conditional probabilities for sample 47000 / 77544
[t-SNE] Computed conditional probabilities for sample 48000 / 77544
[t-SNE] Computed conditional probabilities for sample 49000 / 77544
[t-SNE] Computed conditional probabilities for sample 50000 / 77544
[t-SNE] Computed conditional probabilities for sample 51000 / 77544
[t-SNE] Computed conditional probabilities for sample 52000 / 77544
[t-SNE] Computed conditional probabilities for sample 53000 / 77544
[t-SNE] Computed conditional probabilities for sample 54000 / 77544
[t-SNE] Computed conditional probabilities for sample 55000 / 77544
[t-SNE] Computed conditional probabilities for sample 56000 / 77544
[t-SNE] Computed conditional probabilities for sample 57000 / 77544
[t-SNE] Computed conditional probabilities for sample 58000 / 77544
[t-SNE] Computed conditional probabilities for sample 59000 / 77544
[t-SNE] Computed conditional probabilities for sample 60000 / 77544
[t-SNE] Computed conditional probabilities for sample 61000 / 77544
[t-SNE] Computed conditional probabilities for sample 62000 / 77544
[t-SNE] Computed conditional probabilities for sample 63000 / 77544
[t-SNE] Computed conditional probabilities for sample 64000 / 77544
[t-SNE] Computed conditional probabilities for sample 65000 / 77544
[t-SNE] Computed conditional probabilities for sample 66000 / 77544
[t-SNE] Computed conditional probabilities for sample 67000 / 77544
[t-SNE] Computed conditional probabilities for sample 68000 / 77544
[t-SNE] Computed conditional probabilities for sample 69000 / 77544
[t-SNE] Computed conditional probabilities for sample 70000 / 77544
[t-SNE] Computed conditional probabilities for sample 71000 / 77544
[t-SNE] Computed conditional probabilities for sample 72000 / 77544
[t-SNE] Computed conditional probabilities for sample 73000 / 77544
[t-SNE] Computed conditional probabilities for sample 74000 / 77544
[t-SNE] Computed conditional probabilities for sample 75000 / 77544
[t-SNE] Computed conditional probabilities for sample 76000 / 77544
[t-SNE] Computed conditional probabilities for sample 77000 / 77544
[t-SNE] Computed conditional probabilities for sample 77544 / 77544
[t-SNE] Mean sigma: 9.815991
[t-SNE] KL divergence after 250 iterations with early exaggeration: 103.542763
[t-SNE] KL divergence after 251 iterations: 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.000000
t-SNE done! Time elapsed: 283.6398947238922 seconds

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f7604fa4e80>

#Applying t-SNE on data with PRI_jet_num > 1

[t-SNE] Computing 91 nearest neighbors...
[t-SNE] Indexed 72543 samples in 0.105s...
[t-SNE] Computed neighbors for 72543 samples in 30.156s...
[t-SNE] Computed conditional probabilities for sample 1000 / 72543
[t-SNE] Computed conditional probabilities for sample 2000 / 72543
[t-SNE] Computed conditional probabilities for sample 3000 / 72543
[t-SNE] Computed conditional probabilities for sample 4000 / 72543
[t-SNE] Computed conditional probabilities for sample 5000 / 72543
[t-SNE] Computed conditional probabilities for sample 6000 / 72543
[t-SNE] Computed conditional probabilities for sample 7000 / 72543
[t-SNE] Computed conditional probabilities for sample 8000 / 72543
[t-SNE] Computed conditional probabilities for sample 9000 / 72543
[t-SNE] Computed conditional probabilities for sample 10000 / 72543
[t-SNE] Computed conditional probabilities for sample 11000 / 72543
[t-SNE] Computed conditional probabilities for sample 12000 / 72543
[t-SNE] Computed conditional probabilities for sample 13000 / 72543
[t-SNE] Computed conditional probabilities for sample 14000 / 72543
[t-SNE] Computed conditional probabilities for sample 15000 / 72543
[t-SNE] Computed conditional probabilities for sample 16000 / 72543
[t-SNE] Computed conditional probabilities for sample 17000 / 72543
[t-SNE] Computed conditional probabilities for sample 18000 / 72543
[t-SNE] Computed conditional probabilities for sample 19000 / 72543
[t-SNE] Computed conditional probabilities for sample 20000 / 72543
[t-SNE] Computed conditional probabilities for sample 21000 / 72543
[t-SNE] Computed conditional probabilities for sample 22000 / 72543
[t-SNE] Computed conditional probabilities for sample 23000 / 72543
[t-SNE] Computed conditional probabilities for sample 24000 / 72543
[t-SNE] Computed conditional probabilities for sample 25000 / 72543
[t-SNE] Computed conditional probabilities for sample 26000 / 72543
[t-SNE] Computed conditional probabilities for sample 27000 / 72543
[t-SNE] Computed conditional probabilities for sample 28000 / 72543
[t-SNE] Computed conditional probabilities for sample 29000 / 72543
[t-SNE] Computed conditional probabilities for sample 30000 / 72543
[t-SNE] Computed conditional probabilities for sample 31000 / 72543
[t-SNE] Computed conditional probabilities for sample 32000 / 72543
[t-SNE] Computed conditional probabilities for sample 33000 / 72543
[t-SNE] Computed conditional probabilities for sample 34000 / 72543
[t-SNE] Computed conditional probabilities for sample 35000 / 72543
[t-SNE] Computed conditional probabilities for sample 36000 / 72543
[t-SNE] Computed conditional probabilities for sample 37000 / 72543
[t-SNE] Computed conditional probabilities for sample 38000 / 72543
[t-SNE] Computed conditional probabilities for sample 39000 / 72543
[t-SNE] Computed conditional probabilities for sample 40000 / 72543
[t-SNE] Computed conditional probabilities for sample 41000 / 72543
[t-SNE] Computed conditional probabilities for sample 42000 / 72543
[t-SNE] Computed conditional probabilities for sample 43000 / 72543
[t-SNE] Computed conditional probabilities for sample 44000 / 72543
[t-SNE] Computed conditional probabilities for sample 45000 / 72543
[t-SNE] Computed conditional probabilities for sample 46000 / 72543
[t-SNE] Computed conditional probabilities for sample 47000 / 72543
[t-SNE] Computed conditional probabilities for sample 48000 / 72543
[t-SNE] Computed conditional probabilities for sample 49000 / 72543
[t-SNE] Computed conditional probabilities for sample 50000 / 72543
[t-SNE] Computed conditional probabilities for sample 51000 / 72543
[t-SNE] Computed conditional probabilities for sample 52000 / 72543
[t-SNE] Computed conditional probabilities for sample 53000 / 72543
[t-SNE] Computed conditional probabilities for sample 54000 / 72543
[t-SNE] Computed conditional probabilities for sample 55000 / 72543
[t-SNE] Computed conditional probabilities for sample 56000 / 72543
[t-SNE] Computed conditional probabilities for sample 57000 / 72543
[t-SNE] Computed conditional probabilities for sample 58000 / 72543
[t-SNE] Computed conditional probabilities for sample 59000 / 72543
[t-SNE] Computed conditional probabilities for sample 60000 / 72543
[t-SNE] Computed conditional probabilities for sample 61000 / 72543
[t-SNE] Computed conditional probabilities for sample 62000 / 72543
[t-SNE] Computed conditional probabilities for sample 63000 / 72543
[t-SNE] Computed conditional probabilities for sample 64000 / 72543
[t-SNE] Computed conditional probabilities for sample 65000 / 72543
[t-SNE] Computed conditional probabilities for sample 66000 / 72543
[t-SNE] Computed conditional probabilities for sample 67000 / 72543
[t-SNE] Computed conditional probabilities for sample 68000 / 72543
[t-SNE] Computed conditional probabilities for sample 69000 / 72543
[t-SNE] Computed conditional probabilities for sample 70000 / 72543
[t-SNE] Computed conditional probabilities for sample 71000 / 72543
[t-SNE] Computed conditional probabilities for sample 72000 / 72543
[t-SNE] Computed conditional probabilities for sample 72543 / 72543
[t-SNE] Mean sigma: 22.508396
[t-SNE] KL divergence after 250 iterations with early exaggeration: 97.970383
[t-SNE] KL divergence after 251 iterations: 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.000000
t-SNE done! Time elapsed: 231.39938879013062 seconds

x

<matplotlib.axes._subplots.AxesSubplot at 0x7f76053142b0>

​

tsne_jet_0_df = pd.DataFrame()

tsne_jet_0_df.head()

tsne_jet_1_df = pd.DataFrame()

tsne_jet_2_df = pd.DataFrame()

tsne_jet_0_df.to_csv('tsne_jet_0.csv', index=False)

dfj0 = df_jet_0.copy()

x = dfj0['Label'].str.replace('b', '0', regex=False)

import umap.umap_ as umap

embedding_jet_0 = reducer.fit_transform(dfj0.drop())

plt.figure(figsize=(6,3))

plt.scatter(umap_jet_0_df['umap_1'], umap_jet_0_df['umap_2'])

umap_jet_0_df = pd.DataFrame()

umap_jet_0_df.head()

umap_jet_0_df.to_csv('umap_jet_0_df.csv', index=False)

​

​

​

xxxxxxxxxx

import math

from xgboost import XGBRegressor

xgb_r = XGBRegressor(n_estimators=300)

x

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:587: FutureWarning:

Series.base is deprecated and will be removed in a future version

[16:50:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

1.529429935891298

#applying xgb regressor on data with PRI_jet_num = 1 to predict weight values

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:587: FutureWarning:

Series.base is deprecated and will be removed in a future version

[16:51:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

0.9447710079389715

#applying xgb regressor on data with PRI_jet_num > 1 to predict weight values

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:587: FutureWarning:

Series.base is deprecated and will be removed in a future version

[16:52:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

0.585320408869103

​

​

xxxxxxxxxx

xxxxxxxxxx

xxxxxxxxxx

#applying linear regression on data with PRI_jet_num = 0 to predict weight values

1.7477650489711272

#applying linear regression on data with PRI_jet_num = 1 to predict weight values

1.0627384334604566

#applying linear regression on data with PRI_jet_num > 1 to predict weight values

0.6567532378011104

​

​

xxxxxxxxxx

scaled_jet_0.shape

(99913, 19)

df_jet_0.columns

Index(['EventId', 'DER_mass_MMC', 'DER_mass_transverse_met_lep',
       'DER_mass_vis', 'DER_pt_h', 'DER_deltar_tau_lep', 'DER_pt_tot',
       'DER_sum_pt', 'DER_pt_ratio_lep_tau', 'DER_met_phi_centrality',
       'PRI_tau_pt', 'PRI_tau_eta', 'PRI_tau_phi', 'PRI_lep_pt', 'PRI_lep_eta',
       'PRI_lep_phi', 'PRI_met', 'PRI_met_phi', 'PRI_met_sumet',
       'PRI_jet_all_pt', 'Weight', 'Label'],
      dtype='object')

x

scaled_jet_0_df['Label'] = df_jet_0['Label'].values

scaled_jet_0_df.columns

Index(['DER_mass_MMC', 'DER_mass_transverse_met_lep', 'DER_mass_vis',
       'DER_pt_h', 'DER_deltar_tau_lep', 'DER_pt_tot', 'DER_sum_pt',
       'DER_pt_ratio_lep_tau', 'DER_met_phi_centrality', 'PRI_tau_pt',
       'PRI_tau_eta', 'PRI_tau_phi', 'PRI_lep_pt', 'PRI_lep_eta',
       'PRI_lep_phi', 'PRI_met', 'PRI_met_phi', 'PRI_met_sumet',
       'PRI_jet_all_pt', 'Weight', 'Label'],
      dtype='object')

x

[[27627  2175]
 [ 3761  6403]]
              precision    recall  f1-score   support

           b       0.88      0.93      0.90     29802
           s       0.75      0.63      0.68     10164

    accuracy                           0.85     39966
   macro avg       0.81      0.78      0.79     39966
weighted avg       0.85      0.85      0.85     39966

X = scaled_jet_0_df.drop([ 'Weight', 'Label'], axis=1)

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:587: FutureWarning:

Series.base is deprecated and will be removed in a future version

[18:04:53] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

1.5294179305501039

​

​

xxxxxxxxxx

test_df = pd.read_csv('test.csv')

xxxxxxxxxx

x

#grouping test data like training data

test_jet_0.columns

Index(['EventId', 'DER_mass_MMC', 'DER_mass_transverse_met_lep',
       'DER_mass_vis', 'DER_pt_h', 'DER_deltaeta_jet_jet', 'DER_mass_jet_jet',
       'DER_prodeta_jet_jet', 'DER_deltar_tau_lep', 'DER_pt_tot', 'DER_sum_pt',
       'DER_pt_ratio_lep_tau', 'DER_met_phi_centrality',
       'DER_lep_eta_centrality', 'PRI_tau_pt', 'PRI_tau_eta', 'PRI_tau_phi',
       'PRI_lep_pt', 'PRI_lep_eta', 'PRI_lep_phi', 'PRI_met', 'PRI_met_phi',
       'PRI_met_sumet', 'PRI_jet_num', 'PRI_jet_leading_pt',
       'PRI_jet_leading_eta', 'PRI_jet_leading_phi', 'PRI_jet_subleading_pt',
       'PRI_jet_subleading_eta', 'PRI_jet_subleading_phi', 'PRI_jet_all_pt'],
      dtype='object')

x

#dropping null columns from data with PRI_jet_num = 1

/home/013729625/dsenv/lib64/python3.6/site-packages/pandas/core/frame.py:4117: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

x

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:587: FutureWarning:

Series.base is deprecated and will be removed in a future version

/home/013729625/dsenv/lib64/python3.6/site-packages/xgboost/core.py:588: FutureWarning:

Series.base is deprecated and will be removed in a future version

[18:30:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
             colsample_bynode=1, colsample_bytree=1, gamma=0,
             importance_type='gain', learning_rate=0.1, max_delta_step=0,
             max_depth=3, min_child_weight=1, missing=None, n_estimators=300,
             n_jobs=1, nthread=None, objective='reg:linear', random_state=0,
             reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,
             silent=None, subsample=1, verbosity=1)

x

x

(220156,)

#predicting weight for test data with PRI_jet_num = 0

[18:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

(220156,)

pred_weight_1.shape

(169716,)

#predicting weight for test data with PRI_jet_num = 0

[18:40:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.

(160128,)

x

x

/home/013729625/dsenv/lib64/python3.6/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

#adding predicted weight column

/home/013729625/dsenv/lib64/python3.6/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

	DER_mass_MMC	DER_mass_transverse_met_lep	DER_mass_vis	DER_pt_h	DER_deltaeta_jet_jet	DER_mass_jet_jet	DER_prodeta_jet_jet	DER_deltar_tau_lep	DER_pt_tot	DER_sum_pt	...	PRI_jet_num	PRI_jet_leading_pt	PRI_jet_leading_eta	PRI_jet_leading_phi	PRI_jet_subleading_pt	PRI_jet_subleading_eta	PRI_jet_subleading_phi	PRI_jet_all_pt	Weight	Label
count	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	...	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000	250000.000000
mean	-49.023079	49.239819	81.181982	57.895962	-708.420675	-601.237051	-709.356603	2.373100	18.917332	158.432217	...	0.979176	-348.329567	-399.254314	-399.259788	-692.381204	-709.121609	-709.118631	73.064591	1.646767	0.342668
std	406.345647	35.344886	40.828691	63.655682	454.480565	657.972302	453.019877	0.782911	22.273494	115.706115	...	0.977426	532.962789	489.338286	489.333883	479.875496	453.384624	453.389017	98.015662	1.875103	0.474603
min	-999.000000	0.000000	6.329000	0.000000	-999.000000	-999.000000	-999.000000	0.208000	0.000000	46.104000	...	0.000000	-999.000000	-999.000000	-999.000000	-999.000000	-999.000000	-999.000000	0.000000	0.001502	0.000000
25%	78.100750	19.241000	59.388750	14.068750	-999.000000	-999.000000	-999.000000	1.810000	2.841000	77.550000	...	0.000000	-999.000000	-999.000000	-999.000000	-999.000000	-999.000000	-999.000000	0.000000	0.018636	0.000000
50%	105.012000	46.524000	73.752000	38.467500	-999.000000	-999.000000	-999.000000	2.491500	12.315500	120.664500	...	1.000000	38.960000	-1.872000	-2.093000	-999.000000	-999.000000	-999.000000	40.512500	1.156188	0.000000
75%	130.606250	73.598000	92.259000	79.169000	0.490000	83.446000	-4.593000	2.961000	27.591000	200.478250	...	2.000000	75.349000	0.433000	0.503000	33.703000	-2.457000	-2.275000	109.933750	2.404128	1.000000
max	1192.026000	690.075000	1349.351000	2834.999000	8.503000	4974.979000	16.690000	5.684000	2834.999000	1852.462000	...	3.000000	1120.573000	4.499000	3.141000	721.456000	4.500000	3.142000	1633.433000	7.822543	1.000000

	DER_mass_MMC	DER_mass_transverse_met_lep	DER_mass_vis	DER_pt_h	DER_deltaeta_jet_jet	DER_mass_jet_jet	DER_prodeta_jet_jet	DER_deltar_tau_lep	DER_pt_tot	DER_sum_pt	...	PRI_met_phi	PRI_met_sumet	PRI_jet_num	PRI_jet_leading_pt	PRI_jet_leading_eta	PRI_jet_leading_phi	PRI_jet_subleading_pt	PRI_jet_subleading_eta	PRI_jet_subleading_phi	PRI_jet_all_pt
0	0.995349	-0.038413	0.027727	0.026647	0.010302	0.013589	0.010314	0.019463	0.006051	0.026359	...	0.000564	0.027486	0.014824	0.016759	0.016302	0.016300	0.010548	0.010305	0.010303	0.023016
1	-0.054560	-0.018467	-0.014592	0.365796	0.145052	0.192638	0.145342	-0.163384	0.167182	0.377842	...	0.001444	0.340174	0.168647	0.142420	0.129454	0.129445	0.150277	0.145148	0.145121	0.352807
2	-0.017486	0.204188	0.688189	-0.032016	-0.036906	-0.051953	-0.036880	0.198528	-0.001672	0.061692	...	-0.001873	0.018460	-0.039027	-0.034266	-0.034288	-0.034284	-0.037454	-0.036887	-0.036881	-0.043294
3	-0.023935	-0.080422	0.399547	-0.011000	-0.030274	-0.032693	-0.030394	0.076708	-0.029584	0.038241	...	0.000258	0.048186	-0.036338	-0.028207	-0.029022	-0.029015	-0.030834	-0.030313	-0.030310	-0.034375
4	0.040622	0.067508	-0.244816	0.295481	-0.250106	-0.305083	-0.250444	-0.197506	0.004540	0.018443	...	-0.000262	0.032832	-0.204747	-0.086430	-0.101232	-0.101224	-0.251974	-0.250173	-0.250133	-0.047927
5	0.016219	0.093117	0.185329	-0.067361	-0.021505	-0.040592	-0.021261	0.226532	0.780652	-0.072879	...	-0.002002	0.050763	-0.003538	-0.070937	-0.069324	-0.069312	-0.019187	-0.021407	-0.021411	0.022604
6	0.000684	0.006224	-0.013188	-0.020556	0.007366	0.009416	0.007389	-0.003352	0.066664	0.006039	...	-0.001856	0.012288	-0.002714	-0.023809	-0.024537	-0.024207	0.007891	0.007248	0.007370	0.003812
7	0.017312	0.022180	0.259510	0.137755	0.040645	0.059896	0.040490	0.197480	-0.519610	-0.111481	...	-0.000429	-0.163281	0.023878	0.077593	0.077580	0.077595	0.037895	0.040571	0.040574	-0.025283
8	0.037111	0.223778	-0.157950	-0.119322	0.163851	0.228257	0.163933	-0.032647	-0.075146	-0.010722	...	-0.013782	-0.043541	-0.119344	-0.412197	-0.426591	-0.426564	0.167048	0.163873	0.163839	-0.057361
9	-0.000234	0.004843	-0.002665	-0.000099	-0.001063	-0.002424	-0.001063	0.001844	0.006747	-0.000745	...	-0.033571	-0.006555	0.003600	0.006860	0.007379	0.007285	-0.001193	-0.001070	-0.001080	-0.001630
10	-0.003144	-0.059963	0.009251	0.019069	-0.000016	0.003380	-0.000024	0.017726	-0.002938	0.009200	...	0.979117	0.004438	-0.013002	-0.025541	-0.027976	-0.028630	0.000532	-0.000025	-0.000160	0.012677
11	0.035328	0.623509	-0.108753	-0.171527	0.051418	0.016196	0.051593	-0.226019	0.127588	-0.197955	...	0.072189	-0.231311	0.140567	0.236743	0.270129	0.270041	0.041431	0.051427	0.051407	-0.279006

	EventId	DER_mass_MMC	DER_mass_transverse_met_lep	DER_mass_vis	DER_pt_h	DER_deltaeta_jet_jet	DER_mass_jet_jet	DER_prodeta_jet_jet	DER_deltar_tau_lep	DER_pt_tot	...	PRI_jet_num	PRI_jet_leading_pt	PRI_jet_leading_eta	PRI_jet_leading_phi	PRI_jet_subleading_pt	PRI_jet_subleading_eta	PRI_jet_subleading_phi	PRI_jet_all_pt	Weight	Label
0	100000	138.470	51.655	97.827	27.980	0.91	124.711	2.666	3.064	41.928	...	2	67.435	2.150	0.444	46.062	1.24	-2.475	113.497	0.002653	1
1	100001	160.937	68.768	103.235	48.146	-999.00	-999.000	-999.000	3.473	2.078	...	1	46.226	0.725	1.158	-999.000	-999.00	-999.000	46.226	2.233584	0
2	100002	-999.000	162.172	125.953	35.635	-999.00	-999.000	-999.000	3.148	9.336	...	1	44.251	2.053	-2.028	-999.000	-999.00	-999.000	44.251	2.347389	0
3	100003	143.905	81.417	80.943	0.414	-999.00	-999.000	-999.000	3.310	0.414	...	0	-999.000	-999.000	-999.000	-999.000	-999.00	-999.000	-0.000	5.446378	0
4	100004	175.864	16.915	134.805	16.405	-999.00	-999.000	-999.000	3.891	16.405	...	0	-999.000	-999.000	-999.000	-999.000	-999.00	-999.000	0.000	6.245333	0

	EventId	DER_mass_MMC	DER_mass_transverse_met_lep	DER_mass_vis	DER_pt_h	DER_deltaeta_jet_jet	DER_mass_jet_jet	DER_prodeta_jet_jet	DER_deltar_tau_lep	DER_pt_tot	...	PRI_met_phi	PRI_met_sumet	PRI_jet_num	PRI_jet_leading_pt	PRI_jet_leading_eta	PRI_jet_leading_phi	PRI_jet_subleading_pt	PRI_jet_subleading_eta	PRI_jet_subleading_phi	PRI_jet_all_pt
0	350000	-999.000	79.589	23.916	3.036	-999.000	-999.000	-999.000	0.903	3.036	...	2.022	98.556	0	-999.000	-999.000	-999.000	-999.000	-999.000	-999.000	-0.000
1	350001	106.398	67.490	87.949	49.994	-999.000	-999.000	-999.000	2.048	2.679	...	-1.138	176.251	1	47.575	-0.553	-0.849	-999.000	-999.000	-999.000	47.575
2	350002	117.794	56.226	96.358	4.137	-999.000	-999.000	-999.000	2.755	4.137	...	-1.868	111.505	0	-999.000	-999.000	-999.000	-999.000	-999.000	-999.000	0.000
3	350003	135.861	30.604	97.288	9.104	-999.000	-999.000	-999.000	2.811	9.104	...	1.172	164.707	0	-999.000	-999.000	-999.000	-999.000	-999.000	-999.000	0.000
4	350004	74.159	82.772	58.731	89.646	1.347	536.663	-0.339	1.028	77.213	...	-0.231	869.614	3	254.085	-1.013	-0.334	185.857	0.335	2.587	599.213